Fig. 4: Evaluation of LC-MS2 data processing and result visualization.

a Comparison of matching scores of DDA (with or without deconvolution, ESI⁺) in the complex standards mixture sample. The deconvolution algorithm could significantly improve the matching scores of chemical candidates in comparison to the non-deconvolved spectra. n = 99 independent compounds, means ± SEM; paired one-tailed Student’s t-test (**, significant, p = 1.5e-7, no adjustment). b Statistical analysis of the compound identification results of the 2nd standards mixture dataset. Compared to the other two workflows, MetaboAnalystR reported a significantly higher compound identification percentage. n = 15 independent datasets, means ± SEM; unpaired one-tailed Student’s t-test without adjustment (MetaboAnalystR vs. MS-DIAL/MS-Finder, significant, p = 0.008; MetaboAnalystR vs. MZmine/SIRIUS, significant, p = 0.0003; MS-DIAL/MS-Finder vs. MZmine/SIRIUS, insignificant, p = 0.14). c Number of identified compounds from the 3rd standards mixture dataset using different MS2 reference libraries. d Results of MS2 spectra-based compound identification for the blood exposomics dataset. e Example mirror plot of 2-Piperidinone illustrating the MS2 spectrum matching pattern. The upper side (blue) represents the user’s input, while the bottom side (red) displays the spectrum from the reference library. When the mouse hovers over the fragments, corresponding information, including m/z, relative intensity, and potential formula, is displayed. All matched fragments are marked with a red diamond at the top.